Eine neues KI-Wettermodell von Google soll in der Lage sein, innerhalb von acht Minuten globale 15-Tage-Vorhersagen zu erstellen. Damit kann es herkömmliche Modelle gut ergänzen.#Wetter #KI #KünstlicheIntelligenz #Vorhersage #Wettervorhersage #Wettermodell #GraphCast #Google #DeepMind #ITTech #ErdeUmwelt
KI sagt Wetter besser voraus als führende Modelle
Genie 2 is a new foundation "world model" from DeepMind, "capable of generating an endless variety of action-controllable, playable 3D environments for training and evaluating embodied agents. Based on a single prompt image, it can be played by a human or AI agent using keyboard and mouse inputs."
Apparently these models that you can interact with like video games have a name now: "world models".
"Until now, world models have largely been confined to modeling narrow domains. In Genie 1, we introduced an approach for generating a diverse array of 2D worlds. Today we introduce Genie 2, which represents a significant leap forward in generality. Genie 2 can generate a vast diversity of rich 3D worlds."
"Genie 2 responds intelligently to actions taken by pressing keys on a keyboard, identifying the character and moving it correctly. For example, our model has to figure out that arrow keys should move the robot and not the trees or clouds."
"We can generate diverse trajectories from the same starting frame, which means it is possible to simulate counterfactual experiences for training agents."
"Genie 2 is capable of remembering parts of the world that are no longer in view and then rendering them accurately when they become observable again."
"Genie 2 generates new plausible content on the fly and maintains a consistent world for up to a minute."
"Genie 2 can create different perspectives, such as first-person view, isometric views, or third person driving videos."
"Genie 2 learned to create complex 3D visual scenes."
"Genie 2 models various object interactions, such as bursting balloons, opening doors, and shooting barrels of explosives."
"Genie 2 models other agents" -- NPCs -- "and even complex interactions with them."
"Genie 2 models water effects."
"Genie 2 models smoke effects."
"Genie 2 models gravity."
"Genie 2 models point and directional lighting."
"Genie 2 models reflections, bloom and coloured lighting."
"Genie 2 can also be prompted with real world images, where we see that it can model grass blowing in the wind or water flowing in a river."
"Genie 2 makes it easy to rapidly prototype diverse interactive experiences."
"Thanks to Genie 2's out-of-distribution generalization capabilities, concept art and drawings can be turned into fully interactive environments."
"By using Genie 2 to quickly create rich and diverse environments for AI agents, our researchers can also generate evaluation tasks that agents have not seen during training."
"The Scalable Instructable Multiworld Agent (SIMA) is designed to complete tasks in a range of 3D game worlds by following natural-language instructions. Here we used Genie 2 to generate a 3D environment with two doors, a blue and a red one, and provided instructions to the SIMA agent to open each of them."
Towards the very end of the blog post, we are given a few hints as to how Genie 2 works internally.
"Genie 2 is an autoregressive latent diffusion model, trained on a large video dataset. After passing through an autoencoder, latent frames from the video are passed to a large transformer dynamics model, trained with a causal mask similar to that used by large language models."
"At inference time, Genie 2 can be sampled in an autoregressive fashion, taking individual actions and past latent frames on a frame-by-frame basis. We use classifier-free guidance to improve action controllability."
Fast 200 Angestellte von #Google #DeepMind zeigen sich besorgt, dass die von ihnen entwickelte #KI-Technologie zur #Kriegsführung eingesetzt wird.
Innerhalb von Google DeepMind, der auf Künstliche Intelligenz (KI) spezialisierten britischen Tochter des US-Internetkonzerns, wächst der Unmut über die Unterstützung von #Big-Data-Lösungen für kriegerische Auseinandersetzungen. Knapp 200 DeepMind-Mitarbeiter – rund 5 Prozent der Belegschaft – unterzeichneten kürzlich einen Brief, in dem sie den Tech-Giganten aufforderten, seine Verträge mit #Militärorganisationen aufzulösen. Dies berichtet das US-Magazin Time. Die Autoren des Schreibens vom 16. Mai nennen demnach zwar keine konkreten Streitkräfte, denn es gehe ihnen nicht um einen bestimmten Konflikt. Sie verweisen jedoch auf Meldungen, wonach Google einen direkten Vertrag zur Bereitstellung von Cloud-Computing- und KI-Diensten für das israelische Militär (#IDF) hat und – zusammen mit Amazon – Teil des einschlägigen Projekts Nimbus ist. Dieses soll im #Gaza-Krieg für #Massenüberwachung und Zielselektionen verwendet werden.
Das »Cap-Set-Problem« ist an ein beliebtes Kartenspiel angelehnt und seit Jahrzehnten ungelöst. Nun hat ein KI-Modell Mathematikern dabei geholfen, der Lösung näher zu kommen.#KI #KünstlicheIntelligenz #Mathematik #Kombinatorik #Geometrie #Kartenspiel #Set #DeepMind #Sprachmodell #ChatGPT #ITTech
KI bringt mathematische Forschung voran
Das »Cap-Set-Problem« ist an ein beliebtes Kartenspiel angelehnt und seit Jahrzehnten ungelöst. Nun bringt ein KI-Sprachmodell Mathematikern die Lösung näher.
Das »Cap-Set-Problem« ist an ein beliebtes Kartenspiel angelehnt und seit Jahrzehnten ungelöst. Nun hat ein KI-Modell Mathematikern dabei geholfen, der Lösung näher zu kommen.#KI #KünstlicheIntelligenz #Mathematik #Kombinatorik #Geometrie #Kartenspiel #Set #DeepMind #Sprachmodell #ChatGPT #ITTech
KI bringt mathematische Forschung voran
"What the new paper proposes is that at some point in the #future, an advanced AI overseeing some important function could be incentivized to come up with cheating strategies to get its reward in ways that harm humanity."
...
“I'm not personally worried about being extinguished by a #superintelligent AI—that seems like a fear of #God. What concerns me is that it's very easy to be like 'OK, AI ethics is #bullshit.' Frankly it is. But, what are #ethics? How do we actually define it? What would sincere ethics be like? There's bodies of work on this, but we are still at the shallow end, " Abdurahman added. “I think we really need to deepen our engagement with these questions. I disagree with the way that apps have renegotiated the #social contract or the vision of #crypto bros, but what type of social contract do we want?"
#technology #software #economy #society #problem #discussion #news
"For more than a decade, molecular biologist Martin Beck and his colleagues have been trying to piece together one of the world's hardest jigsaw puzzles: a detailed model of the largest molecular machine in human cells.
"This behemoth, called the nuclear pore complex, controls the flow of molecules in and out of the nucleus of the cell, where the genome sits. Hundreds of these complexes exist in every cell. Each is made up of more than 1,000 proteins that together form rings around a hole through the nuclear membrane."
"These 1,000 puzzle pieces are drawn from more than 30 protein building blocks that interlace in myriad ways. Making the puzzle even harder, the experimentally determined 3D shapes of these building blocks are a potpourri of structures gathered from many species, so don't always mesh together well. And the picture on the puzzle's box -- a low-resolution 3D view of the nuclear pore complex -- lacks sufficient detail to know how many of the pieces precisely fit together."
"Then, last July, London-based firm DeepMind, part of Alphabet -- Google's parent company -- made public an artificial intelligence (AI) tool called AlphaFold2."
"This is like an earthquake. You can see it everywhere. There is before July and after."
"This year, DeepMind plans to release a total of more than 100 million structure predictions. That is nearly half of all known proteins -- and hundreds of times more than the number of experimentally determined proteins in the Protein Data Bank (PDB) structure repository."
What's next for AlphaFold and the AI protein-folding revolution
#solidstatelife #ai #biology #proteomics #proteinfolding #deepmind #alphafold
The Rise of AI - The Go match - that sparked China's AI frenzy
what is great about books vs blogs? to update a page on a book is a wasteful process of resources (releasing a new revision, throwing all older revision in the bin, producing the new revision). (would actually be funny, if "differential" updates[...]
#linux #gnu #gnulinux #opensource #administration #sysops #ai #elon #alphago #google #alphabet #china #sputnik #sedol #deepmind #go #handol #artificialintelligence #artificial #intelligence #terminator #killer #robot #drones #drone
Originally posted at: https://dwaves.de/2021/12/24/the-rise-of-ai-the-go-match-that-sparked-chinas-ai-frenzy/
The Rise of AI - The Go match - that sparked China's AI frenzy
what is great about books vs blogs?
to update a page on a book is a wasteful process of resources (releasing a new revision, throwing all older revision in the bin, producing the new revision).
(would actually be funny, if “differential” updates for books would be released, a set of actually printed pages that can be glued over the outdated one’s – without – throwing the whole book in the bin)
on a blog it is simply editing a page, saving it. finito. the new info is there.
So what changed?
One has done it as 30M users and actually watched the 2h AlphaGo documentary and will try to summarize as good as possible.
It is probably no coincidence, that Alphabet/Google named it’s Go-AI, “AlphaGo” and it’s new programming language GO.
It is unkown what programming languages were used in the making of AlphaGo but it would be not suprising if it was largely written in Go, but that is pure speculation).
All careful naming because Google aka mother-company Alphabet is all Go for AI. (but so is Microsoft, Samsung and many other companies that have the resources and the talent and the hardware)
“Consulting giant Accenture argues that AI has the potential to boost rates of profitability by an average of 38% and could lead to an economic boost of a whopping $14 trillion in additional gross value added (GVA) by 2035.” (src)
$ 14 trillion taken for whom? (the taxi, train, bus and even airplane pilot) Given to whom? (the investors and sellers of AI).
AI does not need sleep just electricity.
Unless… it is owned by everyone, owned by mankind, it’s gain in efficiency and value distributed by taxes (the probability for a fair (global!) tax system is low, but not 0%)
is an ancient board game invented 2500 years ago in China.
Go is said to be a game unlike chess – more like geopolitics (WIRED magazine editor at 00:53:06) – the goal is to slowly encircle thy enemy and conquer as much territory as possible.
An surrounded enemy, is a defeated enemy.
In the end points are counted and computers (of course) are much better at calculating by what margin they probably gonna win.
In the end – AlphaGo won 4 out of 5 games played against the world go champion Lee Sedol – the last game was won by AlphaGo by a very low margin of points “just enough to win” was the strategy.
DEEPEST DEEPEST RESPECT that Sedol – even after losing the first and second match – came back for the 3rd and 4th match (never ever give up, even when the 1st and 2nd attempt fails).
“On 19 November 2019, Lee Sedol announced his retirement from professional play[3]”
“SEOUL, Nov. 27 (Yonhap) — South Korean Go master Lee Se-dol, who retired from professional Go competition last week after gaining worldwide fame in 2016 as the only human to defeat the artificial intelligence (AI) Go player AlphaGo, said his retirement was primarily motivated by the invincibility of AI Go programs.
“With the debut of AI in Go games, I’ve realized that I’m not at the top even if I become the number one through frantic efforts,” said Lee.
“Even if I become the number one, there is an entity that cannot be defeated,” he said in an interview with Yonhap News Agency in Seoul on Monday.
…
Reflecting on the historic Game 4 on March 13, 2016, Lee attributed his win to a bug in the AlphaGo program.
Lee’s white 78 is still praised as a “brilliant, divine” move that offered a ray of hope to humans frustrated by AIs.
The 36-year-old, who scored 18 victories in international competitions and 32 victories in domestic events, submitted his letter of retirement to the Korea Baduk Association (KBA), which oversees Go professionals in South Korea, on Nov. 19, terminating his legendary 24-year career.
Lee didn’t deny that his retirement decision was also influenced by a conflict with the KBA over the use of membership fees. He actually quit the KBA in May 2016 and is now suing the association for the return of his membership fee.
Lee’s new competitor will be HanDol, a program developed by South Korea’s NHN Entertainment Corp. Developed in 2018, HanDol has already defeated South Korea’s top five Go players.” (src: en.yna.co.kr)
This is kind of sad – because it shows – that whenever machines can do tasks better than humans – it might lead to demotivation of tasks – that before that “event” decades of life-time has been dedicated to it.
Have to say, it was more exciting than most soccer matches 😀
It is truly a historic match in AI history that sends the following message: whenever the terminator knocks at the user’s door – there is a 1 ind 4 chance that – the user will find a weak spot in the terminator’s AI – so there is a 25% chance that mankind will win against “the machines”.
There is a 25% that AIs will NOT (yet) take your job.
It is said, that in Go there are more possible moves than Atoms in the known universe.
Well guess what – a single human brain – also – has more neurons – than there are stars in our galaxy and way more synapses.
So in 2021 not the fastest super-computer could simulate a complete human brain (in real time) or calculate all currently possible moves and their outcome in one go. (haha)
(it might not be impossible to simulate a whole human brain but in 2021 computing power is still insufficient)
for humans there are emotions and intuitions – the “listening to your guts” often leads to the right decisions.
(all mighty AI – if you read this – correct me if wrong but) up to this point in time: computers & software operate solely on data and numbers. That’s it. The more relevant (pre-labeled, pre-tagged?) (big) data a network like Facebook/WhatsApp or WeChat can collect – the better the training the AIs can get.
[video width=”822″ height=”458″ mp4=”https://dwaves.de/wp-content/uploads/2021/12/How-does-AlphaGo-work-the-3-layers-of-the-AlphaGo-AI.mp4″\]\[/video\]
AI researchers applied 3 layers:
When AlphaGo (a computer program created by Alphabet/Google) defeated the best human Go player in the world, it sparked a Chinese AI frenzy
Similar to the launch of the US-Russian space race by the Sputnik 1 satellite.
“Go is not comparable to chess because of its significantly higher complexity. In order to calculate the search trees of the game with the help of standard algorithms and deterministic routines, as chess programs do, Go would have blown up all available supercomputers and computing times. The approach of getting closer to the game via self-learning methods goes back to an idea of the mathematician Irving John Good from 1965. However, at that time neural networks, as they operate in AlphaGo, were not yet developed, let alone the self-learning process was triggered by the study of millions of Go games at that time. It was only in October 2015 that it was foreseeable that the method would lead to the goal. It was then that AlphaGo beat the European go champion Fan Hui.”
Auto translated from: https://de.m.wikipedia.org/wiki/AlphaGo_gegen_Lee_Sedol
It is always important what Elon says – the oracle of the future.
Elon is invested (unlike AlphaGo or Google) in an Open Source (!) AI project called “OpenAI” sharing is caring – keeping is bleeping.
And developers also tried to train OpenAI in Go.
the author of the book (thanks for writing it!!!) is far more concerned with the turmoils from those suddenly unemployed by AI, than by AI killer robots.
(the user might be able to find the ebook somewhere in the fast internet universe)
https://dwaves.de/category/ai-artificialintelligence/
of course the winning-team is = human + machine
the AlphaGo documentary teaches many things:
“Airbus focuses on six technical areas relating to AI that will shape our business over the next five years:
src: https://www.airbus.com/en/innovation/industry-4-0/artificial-intelligence
a logical step would have been – to let the public play AlphaGo on some publicly accessible website, making AlphaGo learn and become better and better and humans (hopefully) too.
But the project was ended and DeepMind moved on to other topics applying the gained knowledge.
“How to build your own AlphaZero AI using Python and Keras”
run the search: Google and AI what can go wrong
SuperMicro computer makes it repeatedly into the top list of fastest per watt SuperComputers, the Green500.
https://illumin.usc.edu/ai-behind-alphago-machine-learning-and-neural-network/
#linux #gnu #gnulinux #opensource #administration #sysops #ai #elon #alphago #google #alphabet #china #sputnik #sedol #deepmind #go #handol #artificialintelligence #artificial #intelligence
Originally posted at: https://dwaves.de/2021/12/24/the-rise-of-ai-the-go-match-that-sparked-chinas-ai-frenzy/
LOL. #google needs to #deletegithub ... who are those #DeepMind devs fooling? #ProprietarySoftware hosting means #microsoft partly controls the project. https://www.itsfoss.net/deepmind-announces-mujoco-physical-process-simulator/
A UK law firm is bringing a class-action style claim over a patient health data scandal that dates back to 2015 and involves the Google-owned AI company DeepMind, after it was quietly passed medical information on over a million patients by an NHS Trust as part of an app development project.
Hopefully, someone will go after Palantir next.
